Relative Sensitivity. Precision, and Discriminatory Power of Sediment Toxicity Tests 



To determine the relative sensitivity of each toxicity end-point, the non-parametric K-W 

 test followed by non-parametric Dunnett's t-test was performed to determine significant 

 differences between test samples and respective controls. The numbers and proportions of 

 samples in which toxicity was significantly higher than in the respective controls are 

 tallied in Table 13. The end-points of R. abronius survival, M. edulis abnormal development, 

 and M. edulis percent survival indicated toxicity in the most samples (> 87%). The end- 

 points of A. abdita survival (flow-through conditions), A. abdita avoidance, R. abronius 

 avoidance, and S. purpuratus echinochrome content indicated the least sensitivity (0 to 7% of 

 the samples were indicated as "toxic" relative to controls). Relative to these end-points, 

 those of D. gyrociliatus egg production and S. purpuratus abnormal development were 

 intermediate in sensitivity. The end-points of mitoses per embryo, mitotic aberrations, and 

 cytologic abnormalities in S. purpuratus were recorded in tests of only five samples and 

 indicated sensitivity to a majority of those samples relative to controls. 



To determine the relative abilities of the toxicity tests to discriminate among sites as 

 sampled with the NS&T Program protocols, the non-parametric K-W test was performed 

 with the mean data from each site for each of nine end-points. Significant differences in 

 toxicity between the sites and respective controls were indicated for only four toxicity end- 

 points (Table 14). Percent survival among R. abronius was low in the sediments from the TB 

 and OA sites, but the non-parametric K-W test did not indicate any significant differences (p 

 = 0.11) between the sampling sites and the Puget Sound Control (Table 25). Also, there were 

 no differences in toxicity among the five sites (p=0.12). Avoidance of the sediments from VA 

 by R. abronius was significantly higher than that for Control sediments. Differences among 

 the five sites were indicated (p=0.057), however, site-specific differences could not be 

 identified by non-parametric S-N-K. No differences between sites and controls were 

 indicated with A. abdita survival or avoidance, however, significantly lower survival was 

 indicated in SP sediments than in the others. M. edulis indicated significantly lower normal 

 development and survival in OA sediments than in the controls and in the other four sites. 

 Percent normal development of S. purpuratus indicated TB sediments were significantly more 

 toxic than the controls and the other sites, whereas echinochrome content indicated that TB 

 was least toxic of the five sites. Egg production in D. gyrociliatus did not indicate any 

 significant differences. 



Results of an ANOVA power analysis of minimum detectable differences are summarized 

 in Table 15. The minimum detectable differences between sites for n=2, n=3, n=4, and n=5 

 stations sampled per site, based upon the data collected in this evaluation, are compared. 

 For the n=3 scenario, which is the standard NS&T Program protocol, the end-points of S. 

 purpuratus echinochrome, A. abdita survival, and M. edulis survival would be expected to 

 detect the smallest differences between sites. The largest of the minimum detectable 

 distances would be for the end-points of R. abronius survival and S. purpuratus abnormal 

 development. 



41 



